Converting 2-way to 3-way

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi All,

So I just got done designing and building a two way bookshelf that is awesome to my ears. It's a 2nd order design that also has very linear phase due to time alignment and attention in the crossover design. It's good enough that I now want to make it full range and use it in my main listening rig.

However, I've designed many bookshelf speakers and a few 3-way as well. While some have been good, this one is really something special to my ears. I'm almost sure if I start from scratch on a new 3-way crossover that I'm going to lose something magical that these do. What are the implications of cascading a new 2nd order high pass filter with the existing mid-bass crossover section? If I do a filter that has a flat pass band will it muddy up my phase and lose the nirvana I'm experiencing with these? Is this common practice?




Sent from my iPhone using Tapatalk
 
Recently my brother asked me to turn his Zaph SR-71 into 3 way speaker and i been thinking the same thing. Actually i found Troels project that used same midrange and based on his crossover began simulating my own. So far it looks good and nothing too complex. Try searching for 3ways which use your midwoofer - should be a good starting point.

The main problem with turning 2 way into 3 way i think is baffle step compensation which eats sensitivity of your 2 way. Unless you are ready to sacrifice sensitivity and just want to add bass section - you will probably want to reduce BSC which could require changes to mid-tweet crossover. If you are lucky - simple unwinding of inductor and reduction in tweeter series resistance may get you a good response. But then crossing bass section to midrange gets more complicated as you have to choose crossover point and the slope which will get you the flat response in upper bass.
And of course having your bass section as separate enclosure helps a lot
 
Last edited:
The 2-way already has full baffle step compensation and will stay in its existing cabinet. I plan on building a separate bass cabinet for the woofer and setting the mid cabinet directly on top. The filter applied to the woofer will have to apply BSC depending on the selected crossover frequency. I can probably get away without much if I cross low and maintain a narrow width. I don't want to degrade sensitivity of the current 2-way as its already full BSC with 15w/8530 yielding ~82dB:(. That being the case, it should be easy to find a woofer that is sensitive enough to not have to reduce system sensitivity any further:)


Sent from my iPad using Tapatalk
 
Why would you reduce sensitivity any further? It's the opposite - with 3 way you should try to regain sensitivity you lost to 2 way design by reducing or eliminating BSC! 82db is very low, i'd try to at least get to 85-88... or it's going to be a strange 3 way. I actually think it's going to be more of a problem finding woofer with such a low sensitivity, most decent ones are 90-88 at least...
 
A super low Qts, high xmax, 30Hz fs, 85dB 5" makes for a pretty strange 3-way mid driver. Just sayin'.

Your questions about cascading vs. starting from scratch are answerable in simulation. Put a HPF ahead of the whole 2-way xover and it interacts with the whole system impedance. Just start over but match the acoustic HP rolloff of the original woofer filter exactly, and it will be the same. The only difference will be at the low end, which you're setting out to change.

Okay maybe you'll have to deal with some losses and maaaaybe there'd be an argument to cascading, but I mean sheesh, reconsider active if you're that serious about not touching anything in the filters. It's going to be messy either way moving up to a passive 3-way filter.
 
Well, the thought is that I really like what I hear in the current 2-way. Since it's a 5.5 inch mid-bass, I need more below 80hz to get loud enough for my main rig. I was thinking of crossing < 200 hz LR2. Kinda more of a passive version of sub and satellite rather than full 3 way I guess.

I see where you were going with your previous comment now. I can indeed regain some sensitivity with the woofer if I cross higher but then will have to account for baffle step. However, I really like the voicing and power response of the 2-way design. The higher I cross, the more chance of not being satisfied with one of the best midrange presentations I've ever heard in a speaker....

I'm OK with 82-83 as I'm not chasing SPL and have enough power to drive it. I originally designed the 2-way as speakers for the office and the SS 15w is pretty inefficient to start.


Sent from my iPad using Tapatalk
 
If it gives you any more confidence, there's a commercial loudspeaker that's quite similar to what is being described here. The YG Kipod 1.
SS15w + SS6600 in a waveguide, mated to a powered sub. It's safe to assume that YG are crossing over below 100hz, as they did with the Anat 2 powered bass modules. Woofers are in fact just 8531s or 8530s as shown by numerous people on this forum.

Don't know how sophisticated your sub is, but I recall tinkering with an LR4 at 70hz before disassembling my Zaph ZD5s. Made a big difference in bass quality because my sub didn't have a line level filter.
 
Thanks guys. I have peerless xxls 830843 and TC epic 10" on hand. I'll start playing with crossing to these subs in test boxes in miniDSP and see if I can f ind a good frequency and slope. Ultimately I don't mind leaving a dsp on the sub cabinet.. However, still unsure how to apply a hpf to the 2-way without using DSP or something built into a plate amp:(. So can I just cascade a passive high pass filter before the mid-woofer crossover without effecting the speakers performance in the hpf passband. My front end is just a diy buffalo III dac so I can't use my pre to do it for me.....


Sent from my iPhone using Tapatalk
 
I've been using small mini-monitors with sub for years, first with an analog/active xover I built, and currently with a Linux DSP-based system. I roll the mids out lr-2 at 100, and the subs in lr4. The mini-monitors already do about lr-2 @ 100, so the result is symmetrical. My subs are dual 10" one each side in separate boxes/locations. I've run them mono and stereo, though there really isn't a lot of difference. Pretty happy with the results, especially since going the DSP route.
 
Last edited:
At this point, some people will cringe but you could fill the cabinet volume if you're trying to reduce the bass response. Just calculate for your 0.707 QTC or whatever bass tuning your looking for. I've put old diving weights in my cabinet volume to test out tunings before. Might be cheaper and simpler than a DSP.
 
However, still unsure how to apply a hpf to the 2-way without using DSP or something built into a plate amp:(. So can I just cascade a passive high pass filter before the mid-woofer crossover without effecting the speakers performance in the hpf passband.
Not easily. All the impedance features you created with your existing filter and cabinet are part of the HPF. You're trying to cross low enough that response around the main xover probably isn't going to change, but you've still got the bass impedance peak(s) in the way, and probably components for that BSC you mentioned too.

Making the bass cabinet undersized and matching an active filter on the sub/woofer to the rolloff as mentioned above might be easier to pull off, if the range between excursion problems and a too-high-Q peak is where you want to cross.
 
Not easily. All the impedance features you created with your existing filter and cabinet are part of the HPF. You're trying to cross low enough that response around the main xover probably isn't going to change, but you've still got the bass impedance peak(s) in the way, and probably components for that BSC you mentioned too.

Making the bass cabinet undersized and matching an active filter on the sub/woofer to the rolloff as mentioned above might be easier to pull off, if the range between excursion problems and a too-high-Q peak is where you want to cross.

Ok, I see your point after simulating my proposal in SoundEasy. Trying to passively roll this speaker off at 100-200Hz using a 2nd order filter monkeys with too much further up. Pretty much left with designing a 3-way from scratch, altering box dimensions or using an active filter....

I took a look at Troel's Ellam 3W since I'm using the 15W and created a new 2 way crossover based on the mid/tweeter crossover of that design. With a little optimization I was able to get decent looking results at a 400Hz crossover:( I could not get satisfactory response and phase anywhere below 400Hz..... I guess on the up side, I gain back a bit of sensitivity... Decisions decisions.... I could just love them for what they are and stop wishing they are something they are not:) They sound so good that it makes me want to see if I can.

Attached response and xo designs mentioned above.
 

Attachments

  • 2W_Crossover.PNG
    2W_Crossover.PNG
    8.2 KB · Views: 133
  • 2W_Response.PNG
    2W_Response.PNG
    12.8 KB · Views: 137
  • 3W_Response.PNG
    3W_Response.PNG
    13.5 KB · Views: 121
  • 3W_Crossover.PNG
    3W_Crossover.PNG
    8.2 KB · Views: 136
At this point, some people will cringe but you could fill the cabinet volume if you're trying to reduce the bass response. Just calculate for your 0.707 QTC or whatever bass tuning your looking for. I've put old diving weights in my cabinet volume to test out tunings before. Might be cheaper and simpler than a DSP.

Thanks for mentioning this. The box is ported and tuned to ~52Hz. I guess I can adjust the box volume and seal it. I ran a sim on this using the drivers measured T/S parameters and came up with the attached results. The ported graph is as it is configured right now. The sealed graph has two traces, the lower excursion is at 4 liters sealed vs 11.5 liters sealed. That makes quite a difference! Frequency response looks fine on both with a 3dB in the 70's.
 

Attachments

  • 100W_Excursion.PNG
    100W_Excursion.PNG
    19.4 KB · Views: 113
  • 100W_Sealed_Excursion.PNG
    100W_Sealed_Excursion.PNG
    21.1 KB · Views: 47
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
You could use the bass unit to postpone the baffle step and gain 6dB sensitivity, if you use the mid baffle step as part of its high pass filter. Make the bass baffle large enough so that it's step happens in sub territory. Use a woofer +6dB more sensitive, remove compensation from the mid, lift the tweeter by +6dB and redo the upper crossover.
 
You could use the bass unit to postpone the baffle step and gain 6dB sensitivity, if you use the mid baffle step as part of its high pass filter. Make the bass baffle large enough so that it's step happens in sub territory. Use a woofer +6dB more sensitive, remove compensation from the mid, lift the tweeter by +6dB and redo the upper crossover.


That's another interesting suggestion. I guess I'd have to have asymmetric electrical slopes between woofer and mid... Wouldn't the transition from 2pi to 4pi for the mid require some compensation to achieve second order acoustic targets? It would need like a first order hpf through the transition and a second below. Likewise the woofer would require 3rd order through transition and the second order after that? Working it out in my head this would have to be 4th order or greater to work reasonably well wouldn't it?


Sent from my iPhone using Tapatalk
 
diyAudio Moderator
Joined 2008
Paid Member
If you went this way, the idea would be to remain in 2pi until some frequency where you let it pan out, for example: the schroeder frequency.

The radiation change won't necessarily behave like a filter as even though the response falls on one axis, the sound has mainly changed location.

Baffles might be looked at like horns, as they share characteristics. Unlike horns, a flat baffle is more forgiving of you letting it go below where it's in control. But above some point it's better to keep the radiation pattern to suit the system you want.

But I'm sure you were planning to do that... my point is that you can merge the pattern with a little give and take when necessary. I noticed you wanted to extend the mid's lower response. I won't assume you want to reconfigure its baffle or add wings. At some point I'd expect you'd want to measure for the electrical crossover, especially once you've configured the baffles for the physical crossover.

If I were at a guessing stage, ie: just brought it inside and want to get it going, I'd consider the mid pattern and draw a line in the sand at some frequency for the crossover then assume no more and go for it. If you still aren't ready to measure you might find listening for a while will tell you if you're running one out of its comfort zone.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.